Add MicroCUDA backend, adaptive output wait, wider native K coverage (improve perf)

Shim
- Adaptive output-event wait: sleep to ~120 us before the predicted
  completion (smoothed per wait site), spin near it, back off when late.
  D4R_SHIM_OUTPUT_POLL_US restores a fixed interval.

- Defer the input-marker wait into the worker thread with VRAM interop
  (D4R_SHIM_DEFER_MARKER, default on), taking the host setup off the
  critical path.

MicroCUDA
- Experimental ZLUDA-free backend in wine_nvcuda_bridge, plus
  tools/microcuda and scripts/build_microcuda.sh. Serves accuracy, fast
  and native kernels; unregistered modules fall back to the reference
  path. Unclassified writes invalidate prepared weights.

M kernels
- SWIN_PREP_IDENTITY_AT: key weight prep by allocation identity
  (MicroCUDA); enc3 tube uses it with 8 slots. ZLUDA ignores it.

- dec2 (gfx1101): SWIN_PACKED_RESIDUAL, byte-identical, ~1.6% faster.

- Opt-in gfx12 hardware fp8 conversion (SWIN_FP8_HWCVT), unverified
  on hardware.

Native K translators
- ctile: handle variants that re-materialize the parameter base (mvlo).

- rewrite_output: check live-in registers, not just outputs.

- build.sh: native input covers all 8 flag variants, output the 4 HDR
  "max" variants; depth-tile skip stays mvhi-only.

Tools
- dump_runner and bench_dlss_kernel harness additions.

kernels/native/rewrite_output.py

@@ -43,6 +43,12 @@ def rewrite(source):
     live_out = definitions & uses_outside
     if live_out != {'r215', 'r217'}:
         raise ValueError(f'tile register contract changed: {sorted(live_out)}')
+    # The replacement passes r3, r4, r7, r8 and f1 by name: the region's inputs must be exactly the ones it was
+    # written for (other flag variants of the kernel reuse this region; the checks above alone only cover outputs).
+    region_text = ''.join(lines[i] for i in sorted(removed))
+    live_in = set(REGISTER.findall(region_text)) - definitions
+    if live_in != {'r3', 'r4', 'r5', 'r6', 'r7', 'r8', 'rd1', 'f1', 'f4'}:
+        raise ValueError(f'tile input contract changed: {sorted(live_in)}')
     replacement = '''L__BB0_124:;
     output_tile_fill(KARGP, (int32_t)r3, (int32_t)r4, (int32_t)r7, (int32_t)(r8 + 1u),
                      f1, (AS3 uint8_t*)sharedColor, (AS3 uint8_t*)sharedDepth);